compare-images-in-directories.sh - finds matching images in two sets of directories under a DSSIM similarity threshold

compare-images-in-directories.sh

#!/bin/bash

# Enhanced version that can use dssim, ssim, or ImageMagick
# Usage: ./enhanced_compare.sh <directory1> <directory2> [threshold] [method]

DEFAULT_THRESHOLD=0.01
DEFAULT_METHOD="auto"

usage() {
    echo "Usage: $0 <directory1> <directory2> [threshold] [method]"
    echo "Methods: auto, dssim, imagemagick, perceptual"
    echo "  auto: automatically choose best available method"
    echo "  dssim: use dssim tool (most accurate)"
    echo "  imagemagick: use ImageMagick compare"
    echo "  perceptual: use perceptual hashing"
    exit 1
}

# Function to check available similarity tools
detect_similarity_tools() {
    local available_tools=()
    
    if command -v dssim &> /dev/null; then
        available_tools+=("dssim")
    fi
    
    if command -v compare &> /dev/null; then
        available_tools+=("imagemagick")
    fi
    
    # Perceptual hashing is always available if ImageMagick is present
    if command -v convert &> /dev/null; then
        available_tools+=("perceptual")
    fi
    
    echo "${available_tools[@]}"
}

# Function to use dssim for comparison
compare_with_dssim() {
    local img1="$1"
    local img2="$2"
    
    # dssim returns 0 for identical images, higher values for more different images
    local result=$(dssim "$img1" "$img2" 2>/dev/null | awk '{print $1}')
    echo "$result"
}

# Parse arguments
DIR1="$1"
DIR2="$2"
THRESHOLD="${3:-$DEFAULT_THRESHOLD}"
METHOD="${4:-$DEFAULT_METHOD}"

if [ $# -lt 2 ]; then
    usage
fi

# Detect available tools
available_tools=($(detect_similarity_tools))

if [ ${#available_tools[@]} -eq 0 ]; then
    echo "Error: No image comparison tools available."
    echo "Please install one of: dssim, imagemagick"
    exit 1
fi

# Choose method
if [ "$METHOD" = "auto" ]; then
    if [[ " ${available_tools[@]} " =~ " dssim " ]]; then
        METHOD="dssim"
    elif [[ " ${available_tools[@]} " =~ " imagemagick " ]]; then
        METHOD="imagemagick"
    else
        METHOD="perceptual"
    fi
fi

echo "Using method: $METHOD"
echo "Available tools: ${available_tools[*]}"

echo "Comparing PNG images using perceptual similarity:"
echo "  Directory 1: $DIR1"
echo "  Directory 2: $DIR2"
echo "  Similarity threshold: $THRESHOLD (lower = more strict)"
echo

# Create temporary directory
TEMP_DIR=$(mktemp -d)
cleanup() {
    rm -rf "$TEMP_DIR"
}
trap cleanup EXIT

# Find all PNG files
PNG_FILES1=($(find "$DIR1" -type f -iname "*.png" | sort))
PNG_FILES2=($(find "$DIR2" -type f -iname "*.png" | sort))

if [ ${#PNG_FILES1[@]} -eq 0 ]; then
    echo "No PNG files found in directory 1"
    exit 0
fi

if [ ${#PNG_FILES2[@]} -eq 0 ]; then
    echo "No PNG files found in directory 2"
    exit 0
fi

echo "Found ${#PNG_FILES1[@]} PNG files in directory 1"
echo "Found ${#PNG_FILES2[@]} PNG files in directory 2"
echo "Starting comparison..."
echo

MATCHES=0
COMPARISONS=0
LOWESTTHRESHOLD=1.0
LOWESTTHRESHOLDMATCH1=""
LOWESTTHRESHOLDMATCH2=""

# Compare each image from dir1 with each image from dir2
for img1 in "${PNG_FILES1[@]}"; do
    img1_name=$(basename "$img1")
    echo "Processing: $img1_name"
    
    for img2 in "${PNG_FILES2[@]}"; do
        img2_name=$(basename "$img2")
        ((COMPARISONS++))
        
        # Show progress for large comparisons
        if [ $((COMPARISONS % 100)) -eq 0 ]; then
            echo "  ...completed $COMPARISONS comparisons"
        fi
        
        # Compare images
        similarity=$(compare_with_dssim "$img1" "$img2")
        
        # Check if similarity is below threshold (more similar)
        if [ -n "$similarity" ]; then
            # Use bc for floating point comparison if available, otherwise use awk
            if command -v bc &> /dev/null; then
                is_similar=$(echo "$similarity <= $THRESHOLD" | bc)
            else
                is_similar=$(awk -v sim="$similarity" -v thr="$THRESHOLD" 'BEGIN {print (sim <= thr) ? 1 : 0}')
            fi

            is_less_lowest=$(echo "$similarity <= $LOWESTTHRESHOLD" | bc)
            if [ "$is_less_lowest" -eq 1 ]; then
                LOWESTTHRESHOLD=$similarity
                LOWESTTHRESHOLDMATCH1="$img1"
                LOWESTTHRESHOLDMATCH2="$img2"

                echo "New lowest threshold match found: $similarity"
                echo "  File 1: $img1"
                echo "  File 2: $img2"
            fi

            if [ "$is_similar" -eq 1 ]; then
                echo ""
                echo "*** SIMILAR IMAGES FOUND ***"
                echo "  File 1: $img1"
                echo "  File 2: $img2"
                echo "  Similarity score: $similarity (threshold: $THRESHOLD)"
                
                # Additional info
                info1=$(identify -format "%wx%h %b" "$img1" 2>/dev/null)
                info2=$(identify -format "%wx%h %b" "$img2" 2>/dev/null)
                echo "  File 1 info: $info1"
                echo "  File 2 info: $info2"
                echo ""
                
                ((MATCHES++))
            fi
        fi
    done
done

echo ""
echo "=== SUMMARY ==="
echo "Total comparisons performed: $COMPARISONS"
echo "Similar image pairs found: $MATCHES"
echo "Similarity threshold used: $THRESHOLD"
echo "PNG files in directory 1: ${#PNG_FILES1[@]}"
echo "PNG files in directory 2: ${#PNG_FILES2[@]}"
echo "Lowest threshold match found: $LOWESTTHRESHOLD"
echo "  File 1: $LOWESTTHRESHOLDMATCH1"
echo "  File 2: $LOWESTTHRESHOLDMATCH2"

if [ $MATCHES -gt 0 ]; then
    echo ""
    echo "Similar images were found!"
    exit 0
else
    echo ""
    echo "No similar images found with the current threshold."
    echo "Try increasing the threshold (e.g., 0.1) for more lenient matching."
    exit 1
fi